Create or edit an image with OpenAI

Two OpenAI steps make pictures on your company's own OpenAI key. Create an image — Draws a picture from words with GPT Image. Edit an image — Changes an image as you describe — all of it, or the part a mask shows. Each gives back a PNG file that later steps can upload, attach or post.

Who can do this

Workspace Admins and Editors, on every plan.

Before you start

An OpenAI connection in Connections — see Connect OpenAI. Some OpenAI accounts must verify their organisation before GPT Image works; OpenAI says so in its reason if yours has not.

Steps

Create an image:

1. Select + where the image should be made. In the step picker, open AI, then OpenAI, and select Create an image.
2. In Connection, choose the connection with your key.
3. Optional: in Model, choose one from the list read from OpenAI. Left empty, the step uses OpenAI's current image model.
4. In Prompt, describe the picture, with values in {{ }} — A clean product photo of a blue A4 paper ream on a white desk.
5. In Size, keep Square (1024 × 1024), or choose Landscape, Portrait or Let the model choose.
6. In Quality, keep Auto, or choose Low, Medium or High — higher costs more and takes longer.
7. Select Run this step. Output shows the image, with Download.

Edit an image:

1. In the step picker, under OpenAI, select Edit an image, and choose the Connection.
2. Under Image, select Add a file and drag in a PNG, JPEG or WebP from an earlier step. See Pass files between steps.
3. Optional: under Mask, drag in a PNG the same size, see-through where the image may change. Without one, the model may change any part.
4. In Prompt, say what to change — Put the Consulace logo on the box.
5. Choose Size and Quality as for Create an image, then select Run this step.

Output

Both steps give file (the image, PNG), contentType, revisedPrompt when OpenAI rewrote the prompt, and usage — model, images, tokensIn, tokensOut and seconds. Output shows them as the usage strip. The image travels with the item: drag it into a later step's File or Attachments, or use {{ $json.file }}.

Good to know

• Each item is a separate image, and a separate cost. Each test run draws again — pin the output once you have an image you like; see Pin a step's output.
• A step may take up to five minutes; high quality and large sizes take longest.
• Images OpenAI will not draw — its content rules — fail with its reason, and are not tried again.

If something goes wrong

Related

• Connect OpenAI
• Ask an AI model with Claude, OpenAI or Gemini
• Pass files between steps
• Pin a step's output